Table 3.

Results from Additional Analyses

Variable	All Sample	Among Users†
	Marginal Effect (Robust SE)	Marginal Effect (Robust SE)
	Use of Cancer Drugs in HOPD or Office	Use of Cancer Drug in HOPD	Number of Cancer Drug Claims	Cancer Drug Spending	Spending on Cancer Care Other Than Drugs	Spending on Any Care
Analysis including markets that had had a 340B hospital since years before the study period
(N = 611,948)		(N = 71,135)
NEW340B indicator‡	−0.04 (0.18)	2.68** (1.04)	−0.10 (0.13)	112 (322)	523* (283)	592 (501)
Analysis using Primary Care Service Area (PCSA) as the market
(N = 797,585)		(N = 95,169)
NEW340B indicator§	−0.26 (0.16)	2.51*** (0.76)	−0.22 (0.14)	−315 (310)	99 (236)	−154 (455)
Using false indicator for presence of 340B hospital
(N = 40,451)		(N = 4,809)
False NEW340B¶	0.12 (0.71)	3.12 (2.07)	−0.07 (0.79)	513 (1,226)	425 (766)	1,192 (1,752)

^†Patients who received any provider‐administered cancer drug.

^‡Dummy indicating observations from a market (hospital referral region) that gained a new 340B hospital in a year when the 340B hospital was available.

^§Dummy indicating observations from a market that newly gained a new 340B hospital/clinic in a year when the 340B hospital/clinic was available.

^¶False indicator of the presence of a 340B hospital in the market (hospital referral region).

* p < 0.10, **p < 0.05, ***p < 0.01. All regressions control for market and year fixed effects, and time‐varying patient and market characteristics; standard errors are accounted for clustering within a market; and spending measures are adjusted to 2013 dollars.

HOPD, hospital outpatient department; SE, standard errors.